Question about KSP and game engines

[Halo_003]

Let's say hypothetically it gets announced that KSP 2.0 will not be produced with Unity. I'm curious if other game engines could be used instead. For example, could they use CryEngine or Unreal Engine? Would these work for all of the physics etc required? Personally, I'd drop $120+ for KSP on CryEngine, I'm just curious if it's even possible in theory. (I know it has a 99.9999% probability of not happening, that's okay.) 

This isn't a suggestion, I'm just curious if other game engines would have worked. If so, why was Unity chosen? 

[Guest]

Unity was probably chosen because it's (or appears to be) easy to slap something together in very little time.  The game then grows on top of that.  Unity is also pretty easy for beginners in game programming to get into.  Other game engines probably would have worked just fine.

Edited January 17, 2016 by regex

[worir4]

I wonder if SQUAD regrets using Unity, after seeing all the problems it causes. 

[Guest]

9 minutes ago, worir4 said:

I wonder if SQUAD regrets using Unity, after seeing all the problems it causes. 

While much maligned (I find that, in software, the more popular something is the more maligned it is), Unity is probably as good as any other engine.  For certain some may have been better at the job of KSP but Unity fits the bill, hits a ton of platforms, and has reasonable performance and requirements.  I'm 99% certain a lot of the lag/bugs in the game that are not directly related purely to physics are a product of early KSP code, especially stuff like resource crawling.

E: I also recall someone opining that code obfuscation in KSP hurts performance, but that could be hearsay.

Edited January 18, 2016 by regex

[Waxing_Kibbous]

I would say given the scope and complexities of the game that they would be better off not using an out of the box engine and instead write their own, but that's a pretty large undertaking.

[Gaarst]

Just now, Waxing_Kibbous said:

I would say given the scope and complexities of the game that they would be better off not using an out of the box engine and instead write their own, but that's a pretty large undertaking.

Custom engine for KSP v2.0, still hoping...

[Evanitis]

1 minute ago, Gaarst said:

KSP v2.0

I read that as...

Kerbal Space Program II. - Gravety Strikes Back

[Guest]

10 minutes ago, Waxing_Kibbous said:

I would say given the scope and complexities of the game that they would be better off not using an out of the box engine and instead write their own, but that's a pretty large undertaking.

KSP never would have got off the ground in that case.  For the next one, if it actually happens?  I don't know...

[linuxgurugamer]

18 minutes ago, Waxing_Kibbous said:

I would say given the scope and complexities of the game that they would be better off not using an out of the box engine and instead write their own, but that's a pretty large undertaking.

I would say given the scope and complexities of the game that they ARE better off using an existing engine, rather than first having to totally reinvent the wheel, having numerous bugs along the way, before even beginning on coding the game itself.

[Waxing_Kibbous]

33 minutes ago, regex said:

KSP never would have got off the ground in that case.  For the next one, if it actually happens?  I don't know...

Yeah, I'm not denying that- at the same time they still had to write a lot of custom code and workarounds to make Unity 4 work. One of the members here and Squad staff, Sal_Vager, has a link to the Godot engine http://www.godotengine.org/projects/godot-engine  in their sig- are they a dev on that? Perhaps. These days writing a game engine isn't so uncommon, pick a language and there is probably a game engine written in it. Anyhow it's all idle speculation On the bright side KSP works! Reasonably well even!

 Back to the OP, why Unity? Could be anything, from the original devs were familiar with it, the licensing was appealing (this can be a big factor for Indie companies), multiplatform...

Quote

I would say given the scope and complexities of the game that they ARE better off using an existing engine, rather than first having to totally reinvent the wheel, having numerous bugs along the way, before even beginning on coding the game itself.

There are pluses and minuses obviously-  using Unity is like having a team of coders working for you which is good, on the negative side bugs that you need fixed may not be their top priority. Features and performance enhancements that could help your particular project may not be their top priority. Version changes can break your code in surprising and unpredictable ways. Having ownership of all the code is both labor intensive yet very powerful, so it's a balance that needs to be weighed. But, honestly it's a moot point- I'm not in the trenches making KSP, and neither are most of us here.

Edited January 18, 2016 by Waxing_Kibbous

[TonyC]

Apparently Unity's physics engine was essential for KSP, and switching to another engine would ruin all the mods, force the devs to rewrite everything, and cross-platform compatibility won't be as good as Unity's.

[problemecium]

I'm with @regex. Neither CryEngine nor Unreal Engine is superior to Unity in any significant way. True, they have both been used to produce a number of AAA games with detailed graphics, while Unity is frequently used for simplistic indie games, and Unreal Engine jumped on the PBR bandwagon earlier than Unity; but Unity has caught up graphically, and the sorts of games that get made aren't so much due to engine limitations as the licensing system. Unity caters to indie developers, while Crytek and Unreal have historically tailored their software toward companies.

So KSP on CryEngine isn't anything I would expect to automatically look or perform better. If KSP 2.0 does get produced, it will probably owe improved graphics and performance to advancement in the way the game itself gets programmed - now that they've done it once, SQUAD know what they're doing and can optimize it more effectively.

[r_rolo1]

Well, the biggest hurdle about any engine for KSP is the literal astronomical dimensions of the measurements in game ( distances, velocities, even masses ). People that are around here for longer surely remember the Old Kraken, that was caused by the limitations of Unity ( at the time ,atleast ) regarding calculations with big numbers ( good old floating point issues ) and that Harv hacked around by forcing a change of referential everytime you change focus ( to ensure the numbers never get too big ).

Given that most engines are not made to deal with big landscapes ( as a side point, nowadays I almost burst to laugh when I see a game proudly presenting their huge expansive world of 13*8 Km .KSP really broadens your horizons  ), I would say that most of the premade engines would give KSP the exact same issues ,so TBH if it was to make KSP in another engine, might as well build a custom one

Edited January 19, 2016 by r_rolo1

[Hannu2]

I fear that KSP 2.0 will not be what I want. Instead I hope that there will be some small hobby projects which try to be as realistic as possible and try not to be famous or suitable for everybody. As accurate model of Solar system as possible (and possibility to use custom settings for planets and orbits too), N-body trajectory modeling, possibility to plan and build space programs and ships (more procedurally), FAR-level aerodynamics, highly optimized self made physics engine, sophisticated trajectory prediction and mission control and executing tools, procedurally generated details on bodies (of course not real surface models of bodies), accurate gravity model of non-spherical bodies and much nerdy micromanagement engineering details which can ruin your mission. Graphics could be simple OpenGL stuff but planets should have 3D surface.

I could even take part of such project and program physics routines. Unfortunately whole game is too laborious and have too much boring UI-things to do.

[Guest]

5 minutes ago, Hannu2 said:

I fear that KSP 2.0 will not be what I want. Instead I hope that there will be some small hobby projects which try to be as realistic as possible and try not to be famous or suitable for everybody. As accurate model of Solar system as possible (and possibility to use custom settings for planets and orbits too), N-body trajectory modeling, possibility to plan and build space programs and ships (more procedurally), FAR-level aerodynamics, highly optimized self made physics engine, sophisticated trajectory prediction and mission control and executing tools, procedurally generated details on bodies (of course not real surface models of bodies), accurate gravity model of non-spherical bodies and much nerdy micromanagement engineering details which can ruin your mission. Graphics could be simple OpenGL stuff but planets should have 3D surface.

Orbiter with a VAB/SPH is what you (and I, for that matter) probably want, although on multiple platforms.  Would also be nice to have something that resembled, even slightly, an actual space program management sim.  RP-0 comes pretty close but it is still marred by the "side quest generator" of KSP's "career" mode.

God, career mode is such a mess...  It always comes back to that.

Anyway, yeah, I would love to see that sort of project come about.  A real indie project that wasn't worried about "ease of access" but laser-focused on getting the details correct.  Realism Overhaul has really shown me the light; I want more.